Tunable subnanometer gap plasmonic metasurfaces

resulting in significant enhancement of the fields,2 potentially leading to novel plasmonic sensors,3-5 lenses,6-7 nanolasers ... A representative fal...
4 downloads 16 Views 3MB Size
Subscriber access provided by University of Florida | Smathers Libraries

Article

Tunable subnanometer gap plasmonic metasurfaces Dennis Doyle, Nicholas Charipar, Christos Argyropoulos, Scott A. Trammell, Rafaela Nita, Jawad Naciri, Dr. Alberto Pique, Joseph B. Herzog, and Jake Fontana ACS Photonics, Just Accepted Manuscript • DOI: 10.1021/acsphotonics.7b01314 • Publication Date (Web): 15 Dec 2017 Downloaded from http://pubs.acs.org on December 16, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Photonics is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 11 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

Tunable subnanometer gap plasmonic metasurfaces Dennis Doyle,1 Nicholas Charipar, Christos Argyropoulos,2 Scott A. Trammell, Rafaela Nita,3 Jawad Naciri, Alberto Piqué, Joseph B. Herzog,4 and Jake Fontana* 1

NREIP Summer Student from University of Pittsburgh ONR Summer Faculty Fellow from the University of Nebraska-Lincoln 3 NRC postdoctoral fellow 4 ONR Summer Faculty Fellow from the University of Arkansas Naval Research Laboratory, 4555 Overlook Ave. SW, Washington, D.C. 20375, U.S.A. 2

KEYWORDS: plasmonics, metasurface, subnanometer gap, charge transfer, molecular junction ABSTRACT The index of refraction governs the flow of light through materials. At visible and near infrared wavelengths the real part of the refractive index is limited to less than 3 for naturally occurring transparent materials, fundamentally restricting applications. Here, we carried out experiments to study the upper limit of the effective refractive index of self-assembled metasurfaces at visible and near-infrared wavelengths. The centimeter-scale metasurfaces were made of a hexagonally close packed (HCP) monolayer of gold nanospheres coated with tunable alkanethiol ligand shells, controlling the interparticle gap from 2.8 to 0.45 nm. In contrast to isolated dimer studies, the macro-scale areas allow for billions of gaps to be simultaneously probed and the HCP symmetry leads to large wavelength shifts in the resonance mode, enabling subnanometer length scale mechanisms to be reproducibly measured in the far-field. We demonstrate for subnanometer gaps, that the optical response of the metasurfaces agrees well with a classical (local) model, with minor nonlocal effects and no clear evidence of ligand-mediated charge transfer at optical frequencies. We determine the effective real part of the refractive index for the metasurfaces has a minimum of 1.0 for green-yellow colors, then quickly reaches a maximum of 5.0 in the reds and remains larger than 3.5 far into the near infrared. We further show changing the terminal group and conjugation of the ligands in the metasurfaces has little effect on the optical properties. These results establish a pragmatic upper bound on the confinement of visible and near infrared light, potentially leading to unique dispersion engineered coatings. Recent advances in the understanding of the interactions between electromagnetic fields and matter has led to the development of many exciting optical properties from plasmonic nanostructures. These structures separated by nanometer size gaps can confine electromagnetic fields into spatial regions well below the diffraction limit,1 resulting in significant enhancement of the fields,2 potentially leading to novel plasmonic sensors,3-5 lenses,6-7 nanolasers,8-10 photovoltaics,11 and nanoscale antennas.12-13 To maximize the confinement of the electric field and ensuing optical properties such as the effective index of refraction,14-15 classical theory predicts as the gap size between two plasmonic nanospheres approaches zero the local field enhancement can become arbitrarily large, which is not physically possible, therefore breakdown of the field must occur.16 The onset for the breakdown of electric field (nonlocal response) in the gap between the nanospheres in vacuum has been shown to experimentally occur at length scales comparable to or less than the atomic lattice constant of gold (0.41 nm).17-18 Other experimental work reported if organic ligand spacers are placed in the gap, that nonlocal19 and full breakdown, allowing charge transfer for gaps as large as 1 nm, can be observed.2021 Yet, open questions remain as to the mechanisms that would facilitate charge transfer through the ligands at optical frequencies and resulting charge-transfer plasmon modes.22 To investigate these subnanometer gap mechanisms, we created large-area metasurfaces using directed selfassembly. The metasurfaces are comprised of a HCP monolayer of gold nanospheres coated with tunable alkanethiol ligand shells, with interparticle gaps approaching atomic lattice length scales. We experimentally show that the optical response of the metasurfaces agrees well with a classical model, with only small nonlocal effects. We demonstrate the effective real part of the refractive index is as small as 1.0 at 555 nm and as large as 5.0 at 743 nm. Finally, we find changes to the optical properties of the metasurfaces are negligible if the terminal group or conjugation of the ligands is varied. RESULTS AND DISCUSSION

1 ACS Paragon Plus Environment

ACS Photonics 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 11

The metasurfaces were fabricated using a robust ‘one-pot’ directed self-assembly technique based on phase separation and surface tension gradients.23-24 Briefly, a suspension of alkane thiol ligands in a water miscible organic solvent was mixed into an aqueous suspension of gold nanospheres. The thiolated ligands rapidly covalently bond to the surface of the gold nanospheres, causing the nanospheres to phase separate to the air-fluid interface, forming macroscale, monolayer metasurfaces. The metasurfaces then self-transfer onto pre-wetted glass substrates by utilizing the surface tension gradients arising from the water-organic solvent mixture (see Methods section for details). The resulting metasurface geometry (Figure 1a) consists of a monolayer of HCP gold nanospheres coated with ligand shells of tunable length on a glass substrate. The interparticle gap is controlled by varying the chain length of the alkane thiol ligands. A mixture of equal length alkane chains with mono- and di-thiols was used. The monothiol ligands bond to the surface of the gold nanospheres leaving the alkane chain exposed, providing a repulsive (entropic) force aiding to prevent aggregation while dithiols provide an attractive force covalently bonding the nanospheres together. The chain lengths were varied from C2 to C14, where the C-number indicates the number of carbon atoms in the alkane chains.

Figure 1. (a) Schematic of the metasurface. (b) False-colored TEM image of a metasurface (C3). The Fourier transform of the image is shown in the upper-right. The average nanosphere diameter, 13.3±0.7 nm, can be used to scale the TEM image. Transmission (c) and reflection (d) true-color photos of a metasurface (C3). The metasurfaces were trimmed into 7 by 7 mm squares using a UV laser ablation system. A representative false-colored TEM image of a metasurface is shown in Figure 1b, validating the metasurfaces are monolayers over macro-length scales while maintaining nanometer sized gaps. The image in the upper-right corner of Figure 1b is the Fourier transform of the TEM image, showing the six-fold HCP symmetry. Photographs of a 0.5 cm2 metasurface, backlit with white light, is shown in transmission (Figure 1c) and reflection (Figure 1d). The gold colored reflection is a consequence of the dense packing of gold nanospheres and the spatially uniform blue color indicates the nanoparticles are not aggregated (touching). To quantify the interparticle gap for each metasurface, transmission electron microscopy (TEM) grids were prepared by gently scooping the metasurfaces with the grids from the air-fluid interface as they were assembling. From the TEM imagery, 1,000 gaps were measured for each metasurface. The nanosphere diameter from all measurements was determined to be 13.3±0.7 nm. The distribution of gap sizes is shown in Table 1 and the histograms in Figure 2. The gaps ranged from 2.8 nm for C14 down to 0.45 nm for C2. The metasurfaces hexgonally close packed at nanoand micro-meter length scales, though the packing was amorphous at larger size scales. The full-width-at-halfmaximum (FWHM) values from the gap histograms in Figure 2 shows the gap uniformity decreased as the ligand length increased. In addition to tuning the interparticle gaps using different ligand lengths, the introduction of the dithiol ligands is a significant and key advancement over previous reports,24 improving the uniformity and area of the metasurfaces (Figure 1) and highlighting the need for both attractive and repulsive forces during the self-assembly process. Moreover, the dithiols also mechanically strengthened the metasurfaces sufficiently to enable reliable transfer to TEM grids, thereby facilitating direct measurements of the interparticle gaps (Figure 2).

2 ACS Paragon Plus Environment

Page 3 of 11 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

Figure 2. False-colored TEM images (top) and interparticle gaps distributions (bottom) for ligand mixtures; C2 (1ethanethiol/1,2-ethanedithiol), C3 (1-propanethiol/1,3-propanedithiol), C6 (1-hexanethiol/1,6-hexanedithiol), C8 (1octanethiol/1,8-octanedithiol), C10 (1-decanethiol/1,10-decanedithiol) and C14 (1-tetradecanethiol/1,14tetradecanedithiol).

‫܌ܖ܉܏ܑܔ‬ C2 C3 C6 C8 C10 C14

Table 1. metasurface gaps gap (nm) ∆gap (nm) 0.45 ±0.14 ±0.22 0.55 ±0.46 1.1 1.4 ±0.37 ±0.48 1.6 ±0.69 2.8

The gap size depends on the number of carbon atoms in the alkane chains, C (Figure 3a, black dots). The interparticle gap has been shown to increase at 0.12C nm for HCP gold nanospheres coated with alkanethiols.25-26 If it is assumed the dithiols ligands set the gap size due to the additional 0.24 nm Au-S bond offset, relative to the monothiol ligand, and are interdigitated, then the gap size is expected to be of the form 0.12C+0.48 nm (Figure 3a, solid black line). We find reasonable agreement between the theoretical prediction and the experimental data in Figure 3a, demonstrating the gap size is predominantly set by the length of the dithiol ligands. As the interparticle gap decreases, the enhancement of the electric field in the gap dramatically increases leading to the absorbance peak wavelength red-shifting due to the increasing coupling strength between the individual nanosphere modes.27 The normalized experimental absorbance peak spectra from each metasurfaces is shown in Figure 3b. The absorbance peak shifts by ~175 nm covering yellow (580 nm for C14) to near infrared red (775 nm for C2) wavelengths. The quality factor (Q=peak wavelength/FWHM) of the absorbance peaks also remains large throughout. Figure 3c relates the absorbance peak wavelength to the interparticle gap. The experimental peaks are represented by the black dots. The uncertainty in the peak wavelength was determined by averaging measurements from 5 different spatial positions (beam area ~ 4 mm2) on 3 separate metasurfaces for each ligand length. The uncertainty in the gap sizes was determined from the TEM analysis (Table 1). The onset for the breakdown of the electric field can be described using a nonlocal hydrodynamic model where the local electric field in the gap is attenuated through the introduction of a β−parameter, which is proportional to the Fermi velocity of the electrons. In the nonlocal model the electron current density, J, inside the metal induced by the applied electric field, E, oscillating at frequency, ω, is given by,19, 28-29 ߚ ଶ ߘሺߘ ∙ ࡶሻ + ሺ߱ଶ + ݅ߛ߱ሻࡶ = ݅߱߱௣ଶ ߝ଴ ࡱ

(1)

3 ACS Paragon Plus Environment

ACS Photonics 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 11

where γ is the dampening coefficient, ωp is the plasma frequency of gold and ε0 is the permittivity of free space (see Methods section for details).

Figure 3. (a) Interparticle gap as a function of the number of carbon atoms, C, in the alkane chains. (b) Normalized experimental absorbance spectra of the metasurfaces for each ligand length. (c) Experimental absorbance peak wavelength (black dots) of the metasurfaces as a function of interparticle gap. Finite element simulations of the metasurface with hexagonally close packing of the nanospheres with a local response (thick solid black line), a nonlocal response (thin dotted black, β=1x106 m/s; thin dotted grey, β=3x106 m/s lines), and local response from isolated dimers (thick solid grey line) as a function of interparticle gap. To understand the experimental absorbance peak shift as the gap decreases three-dimensional finite-element simulations (COMSOL Multiphysics 5.3) of HCP metasurfaces with a classical (local) and nonlocal response were carried out (see Methods section for details). The thick solid black line in Figure 3c shows the model results of the HCP metasurfaces with a local response, setting an upper bound on the field enhancements and absorbance peak red-shift. Equation 1 was directly incorporated into the local HCP metasurface model, introducing the nonlocal effects. The nonlocal response is shown in Figure 3c. Two values of β were chosen; β=1x106 m/s (thin dotted black line) and β=3x106 m/s (thin dotted grey line). The nonlocal effects retrieved from the simulations for β=1x106 m/s start to become appreciable for gap sizes less than ~1 nm. As β increases to 3x106 m/s, the electric field inside the gap is further attenuated, leading to a reduction in the absorbance peak wavelength shift. The experimental data is in good agreement with the nonlocal simulations for β=1x106 m/s.22 On the contrary, when β=3x106 m/s the simulated data deviates significantly from the experimental results, especially for the subnanometer gap regime where nonlocal effects are expected to be triggered. The local response of an isolated dimer was also modeled in Figure 3c (thick solid grey line). The results show the significance of the six-fold near-neighbor coupling from the HCP symmetry, relative to the single near-neighbor in the dimer geometry, needed to red-shift the absorbance peak far enough to overlap with the experimental data. The dimer evolution therefore places a lower bound on the absorbance peak shift. For the larger gap sizes, >1 nm (C8-C14), the absorbance peak wavelength is blue-shifted relative to the local HCP metasurface model upper bound, yet shifted well above the local dimer lower bound. Since the nonlocal effects are negligible for these gap ranges22 it is assumed the reason is from non-ideal experimental HCP structures. This assumption is supported by the data in Figure 2, showing the distribution of the gap sizes increases as the ligand length become longer. Note that the blue-shift in the experimental data cannot be attributed to quantum effects due to the nanoparticle size, since these type of quantum size effects start to appear for nanoparticles with diameters less than 8 nm.18

4 ACS Paragon Plus Environment

Page 5 of 11 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

As the gap decreases below 1 nm (C6-C2), the nonlocal models can theoretically deviate significantly from the local model depending on the value of β. While the nonlocal effects are present, the maximum difference between the local model and the experimental data is only 14 nm for C2, less than a 2% difference. The spectroscopic ellipsometry parameters Ψ and ∆ for the C6 metasurface are reported in Figure 4a and 4b. The effective real part of the refractive index of the C6 metasurface retrieved from the ellipsometry parameters is shown in Figure 4c (see Methods section for details). The real part of the refractive index shows a minimum of 1.0 at 555 nm, a maximum of 5.0 at 743 nm and remains larger than 3.5 out to 1,550 nm. The strong resonating response and large local field enhancement of the metasurface are the physical reasons for the large real part of the computed effective refractive index. Both the magnitude and dispersion of the refractive index is large for these engineered metasurfaces, surpassing that of any known naturally occurring transparent material at visible and near infrared frequencies.14-15, 24

Figure 4. Representative ellipsometry parameters Ψ (a) and ∆ (b) for the C6 metasurface (solid black lines), the grey area is the experimental uncertainty and the dashed red lines are the 3-layer model fits. Real (c) and imaginary (d) parts of the effective refractive index for the C6 metasurface. Real (e) and imaginary (f) parts of the effective refractive index of the metasurfaces as a function of ligand length. Figure 4e shows the evolution of n as a function of gap. As expected, the spectra red-shifts in wavelength as the length of the ligand decreases from C14 to C2. The largest (5.0) refractive index occurs for a gap of 1.1 nm (C6) and not at the smallest gap of 0.45 nm (C2) as expected if the response was purely local. From Figure 2, C6 has a larger gap size distribution, relative to C3 and C2, implying a lower quality HCP structure, consequently the extreme enhancement for C6 is not due to a reduction in packing efficiency for gap sizes below 1 nm. The C6 gap length scale is consistent with the established local-to-nonlocal crossover regime, ~1 nm.19 Therefore, a plausible explanation for the extreme enhancement for the C6 metasurface is that an upper bound has been reached for the electric field enhancement, and resulting refractive index, before nonlocal effects become noticeable, in agreement with Figure 3. The effective imaginary part of the refractive index of a metasurface (C6) retrieved from the ellipsometry measurements is shown in Figure 4d. The peak maxima occurs at 629 nm with a quality factor of 4.3. The figure of merit, n/k, at the peak maxima peak is 0.7. However, the imaginary part of the refractive index drops close to zero approaching near infrared wavelengths, while the real part of the refractive index remains large. As a consequence,

5 ACS Paragon Plus Environment

ACS Photonics 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 11

the figure of merit is 28.5 at 1,550 nm, forty times larger than at visible wavelengths, which may have useful telecommunication application consequences.30 Figure 4f shows the peak of the imaginary part of the refractive index red-shifting as the gap decreases, in agreement with absorbance data collected in Figure 3c.27 Subtle differences in ligand composition are known to dramatically influence the electrical properties through the ligand at low frequencies, e.g. the electrical conductivity decreases by four orders of magnitude if di-thiol alkane ligands are replaced with mono-thiols.31 An unresolved issue is the effect ligands have at higher frequencies on the optical response.22 To determine how the terminal ends of the ligands influence the optical response of the metasurfaces, ligands with alkane or thiol end groups were exchanged. The ligand shells were composed solely of mono- (1-propanethiol (C3mono)) or di-thiols (1,3-propanedithiol (C3-di)) and compared to the standard mixed ligand shells (1propanethiol/1,3-propanedithiol (C3)). Furthermore, to understand the role of ligand conjugation, the alkane ligand cores were replaced with a highly conjugated core; 1,4-benzenedithiol (BZT). Figure 5a shows the experimental absorbance spectra for the C3, C3-di, C3-mono and BZT metasurfaces. Unsurprisingly, the C3-di absorbance spectrum closely matches the C3 spectrum, since the di-thiol ligands set the gap size, with similar peak wavelengths and quality factors, see Table 2.

Figure 5. (a) Normalized experimental absorbance spectra for the C3, C3-di, C3-mono and BZT metasurfaces. The grey area is the uncertainty. Effective real (b) and imaginary (c) refractive index of the BZT metasurface.

ligand C3 C3 − di C3 − mono BZT

Table 2. metasurface ligands abs. peak (nm) FWHM (nm) 724 191 728 191 702 168 710 215

Q 3.8 3.8 4.1 3.3

For the C3-mono case in Figure 5a, the peak wavelength is slightly blue shifted by 3% (22 nm) relative to the C3 peak, which is expected if tunneling is occurring. However, the quality factor for C3-mono is larger (4.1) compared to the C3 metasurface (3.8), which is not expected if tunneling is occurring, since it would broaden the C3-mono peak and decreases the quality factor less than C3.22, 28, 32-33 Moreover, it is well established that mono-thiols are significantly more insulating than di-thiols at low frequencies in molecular junctions.31 This relationship is expected to hold at elevated frequencies.20 Entropic forces are anticipated to dictate the packing of the C3-mono nanospheres during the metasurface assembly process, due to the absence of the additional thiol. Therefore, it is reasonable the C3-mono ligands are not interdigitated, as in the C3 and C3-di cases, leading to the gap being approximately twice the ligand length. The C3 gap was experimentally determined to be 0.55±0.22 nm (Table 1). If it is assumed the C3-mono gap is twice the C3 gap or approximately 1 nm, then the absorbance peak wavelength is in reasonable agreement to the local response

6 ACS Paragon Plus Environment

Page 7 of 11 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

model in Figure 3c. If the gap is assumed to be only one ligand length in size, or approximately 0.6 nm, it would correspond closely to the β=3x106 m/s evolution, in disagreement with all the other experimental evidence. Significant effort was made to transfer the C3-mono metasurfaces to TEM grids to enable direct imaging of the gaps. Due to a lack of an attractive di-thiol mechanism between the nanospheres during the self-assembly process, the C3-mono nanospheres packed very poorly on the TEM, leading to unreliable gap distribution measurements. Without the availability of reliable TEM imagery direct conclusions cannot be drawn on the uniformity of the hexagonally close packing, yet if the packing were not uniform, the quality factor would be expected to be smaller than the C3 metasurfaces. This indirect evidence implies the C3-mono metasurface are well packed with uniform gaps. Based on all the evidence above, the C3-mono metasurfaces are unlikely to be transferring charge at optical frequencies. The BZT metasurface absorbance spectrum is shown in Figure 5a. The BZT absorbance peak wavelength is blue shifted by 2% (14 nm) relative to the C3 metasurface, but red shifted by 1% (8 nm) compared to the C3-mono metasurface (Table 2), which falls within the experimental uncertainty window. The quality factor for the BZT metasurface is 13% smaller than the C3 and 20% smaller than the C3-mono metasurfaces, due to an increased FWHM. The increased FWHM can be a consequence of non-uniform HCP packing, or it may also be a consequence of tunneling.22, 28, 32-33 As in the C3-mono case, reliable gap distribution measurements were not possible for the BZT metasurfaces due to inability to transfer the metasurfaces well onto the TEM grids. To unravel the origins of the increased FWHM for the BZT metasurface the effective real and imaginary refractive indices were determined from spectroscopic ellipsometry measurements and are shown in Figure 5b and 5c, respectively. A previous experimental report demonstrated that if charge is transferred between nanoparticles interconnected with BZT, then a secondary near infrared absorbance peak emerged.20 Theoretical investigations for nanospheres bridged with BZT also showed the primary absorbance peak blue shifts and broadens and a secondary near infrared absorbance peak can emerge, but is highly dependent on the conductivity of the molecular junction and geometry of the nanoparticles.22 From the imaginary part of the refractive index in Figure 5b we find no evidence of a secondary near infrared absorbance peak as far out as 1550 nm. Based on the minimal absorbance peak blue shift and slight broadening, as well as the lack of an additional absorbance peak, these results suggest the BZT metasurfaces are also not likely to be transferring charge at optical frequencies and the increased FWHM is a consequence of non-uniform nanosphere packing. In addition, direct quantum tunneling effects leading to charge transport are possible only for very small gaps (